Replication package: Why do sites disagree? Attributing coefficient displacement to an availability mechanism from site-level regression summaries

Code, aggregate outputs, figures and documentation for the article "Why do sites disagree? Attributing coefficient displacement to an availability mechanism from site-level regression summaries", submitted to Biometrical Journal. Contains the detection, attribution and repair procedure for site-level regression summaries, the simulation study, the retail panel and NHANES applications, the retail date audit, and the registration files for the 2021-2023 NHANES check. NHANES files are not included; code/nhanes-fetch.py downloads them from the National Center for Health Statistics. Full run: python code/run-all.py (about one minute). See README.md.